Decipher mesoscale chemical complexity in high entropy alloys by scaling Monte Carlo simulation to extreme

Mesoscale chemical defects, such as precipitates, are vital for the superb mechanical and chemical properties of high entropy alloys, but present a challenge for tradtional atomistic simulation methods due to their limitations in spatial and temporal scale. Here we demonstrate that the recently proposed SMC-X (scalable Monte Carlo at eXtreme) method can overcome this long-standing problem. SMC-X is a generalized checkerboard algorithm that enables highly efficient parallel MC moves for arbitrary short-range interactions, including machine learning models. By coupling SMC-X, DFT data, ML model, and high-performance computing, we demonstrate that unprecedented atomistic spatial and temporal scales can be reached, with near-DFT accuracy. This breakthrough enables us to directly observe the formation of mesoscale chemical defects, shedding light on simulation-guided design of chemically complex materials.